Strategies for effective Web services adoption for dynamic e-businesses

Decision Support Systems 42 (2006) 789 – 809 www.elsevier.com/locate/dsw

Strategies for effective Web services adoption for dynamic e-businesses Andrew N.K. Chen*, Sagnika Sen, Benjamin B.M. Shao Department of Information Systems, W. P. Carey School of Business, Arizona State University, Tempe, AZ 85287-4606, USA Received 10 August 2004; received in revised form 9 May 2005; accepted 11 May 2005 Available online 17 June 2005

Abstract Web services hold the promise for the so-called dynamic e-business movement. Currently, many organizations are either in the process of adopting Web services technology or seriously evaluating this option. One of the major concerns of senior management in this endeavor is the cost of adopting Web services. In this paper, a model is proposed to evaluate an organization’s position in a technology adoption space by evaluating its current level of information technology (IT) sophistication. The model identifies critical factors necessary for the successful adoption of Web services technology along three dimensions—intranet, extranet, and Internet. A simulation experiment is conducted to find the most cost-effective strategy for allocating resources to pursue Web services adoption. Alternative strategies are evaluated under three scenarios with different combinations of significance levels (weights) and diffusion levels of the critical factors. Our results suggest that different strategies should be employed, while organizations consider their existing organizational IT status and focus area. This study provides useful guidelines for management to utilize available resources effectively in the process of adopting Web services technology. D 2005 Elsevier B.V. All rights reserved. Keywords: IT sophistication; Technology adoption and impact; Web services; Electronic business

1. Introduction E-business applications are now commonplace in business-to-consumer (B2C) and business-to-business (B2B) environments. Lately, a movement towards the so-called dynamic e-business has been gathering momentum and it has been claimed to be the next stage * Corresponding author. Tel.: +1 480 965 2687; fax: +1 480 965 8392. E-mail address: [email protected] (A.N.K. Chen). 0167-9236/$ - see front matter D 2005 Elsevier B.V. All rights reserved. doi:10.1016/j.dss.2005.05.011

of e-business. Dynamic e-business is concerned with how organizations can integrate systems across intranets, extranets, and the Internet in a dynamic fashion, permitting them to modify the existing systems quickly and easily when the business process requires some changes. Dynamic e-business is defined as the next generation of e-business that focuses on the integration and infrastructure complexities of B2B by leveraging the benefits of Internet standards and common infrastructure to produce optimal efficiencies for intraand inter-enterprise computing [29]. Maruyama [44]

790

A.N.K. Chen et al. / Decision Support Systems 42 (2006) 789–809

referred to dynamic e-business as the third wave of ebusiness evolution after B2C and B2B. Equipped with advanced Internet technologies and standards, participants in dynamic e-business are able to externalize a company’s business processes in a standard way and utilize business processes provided by other parties to create new applications or business flows by integrating such internal and external processes dynamically. Web services hold the promise for this dynamic ebusiness movement. The Web services paradigm has emerged as an important mechanism for interoperation among separately developed and geographically distributed applications [46]. Web services describe a family of standards that allow participants of e-business to discover each other (e.g., using Universal Description, Discovery, and Integration—UDDI), describe their applications/services (e.g., using Web Services Description Language—WSDL), establish communication based on familiar protocols (e.g., using Hypertext Transfer Protocol—HTTP), issue requests and responses (e.g., using Simple Object Access Protocol—SOAP), and exchange data and information (e.g., using eXtensible Markup Language—XML). Web services enable individuals and organizations to design, implement, deploy, and use the so-called plug-and-play applications, both internally and externally, in a ubiquitous fashion. In a similar vein, companies can use standard platformindependent Internet technologies offered by Web services for three types of integration—internal, external, and multi-channel [1,2]. For internal implementation (or application-to-application interaction), integration of heterogeneous systems can be achieved at lower cost and enterprise systems can be exposed as a set of reusable Web services that can be consumed by composite business applications. For external implementation (or inter-enterprise connectivity), business partners can integrate transactions based on agreed Web services standards for each step in a business process so as to reduce custom business processes. For multi-channel implementation (or global access by extended users over the Web), extended integration across parties of value and supply chains can re-tool their existing applications across different channels to adapt to business innovations and extend the reach of systems. Both the academic and practitioner literature has recognized the immense potential of Web services in

providing interoperability among heterogeneous information systems. While there is no lack of articles in popular press and academic journals on Web services technology, architecture, and standards, there has been little research effort to address the specific issue of Web services adoption in an organization. Adoptionrelated issues from a business information technology perspective–such as the most cost-effective strategy to implement this technology within an organization and the driving factors for successful implementation– have been under-addressed so far. Implementation of any new technology requires a careful assessment of the needs and capabilities of the organization as well as the formulation of cost-effective adoption strategies. Research efforts are thus needed to investigate the adoption decision process in the Web services context. We argue that the first step in the adoption of a new technology such as Web services involves a proper assessment of the status of information technology (IT) within the organization. Based on the current IT level and resource availability, appropriate strategies can be formulated. Hence, in this paper, we first attempt to formalize a way to locate the position of a company in an IT adoption space based on the bsophistication levelQ of its current IT infrastructure, factors, and applications. We extend the adoption space model proposed by Chen et al. [17] to measure the resources that a company would need in order to adopt Web services applications with consideration of the company’s current IT status along three dimensions of Web-based applications—intranet, extranet, and Internet. Our research aims to help businesses in the adoption decision of Web services, and our simulation results provide guidelines for businesses with limited resources to pursue the adoption of Web services in a cost-effective manner by strengthening appropriate factors. The contribution of this study is two-fold. First, we identify the factors relevant to the adoption of Web services by an organization and provide an estimate of the potential resources that would be needed by the adoption. In introducing any new technology, one of the major concerns of senior management is the resources associated with the change. The success of such an endeavor depends on various factors. We argue that the higher the level of IT sophistication of a company, the less the resources needed to adopt Web services technology. Hence, we first identify


the factors that collectively determine the level of IT sophistication of a company. We can then measure the status of IT sophistication by considering (1) the relative importance of the factors and (2) their current level of diffusion within the specific organization. The second contribution is that we provide simulation results to draw useful guidelines for management to effectively utilize available resources in the process of adopting Web services technology. Our results suggest that different strategies should be employed while organizations consider their existing IT sophistication and focus area. Allocating resources on an ad hoc basis to the weakest areas may not always lead to the best utilization of available resources. The rest of the paper proceeds as follows. The next section, according to relevant literature, introduces important organizational and technological factors contributing to IT sophistication and influencing the adoption of Web services. Section 3 presents the model and discusses research goal and research design. The simulation results of the main experiment and sensitivity analysis for examining effective Web services adoption are provided in Section 4. We follow with result discussions and managerial implications in Section 5 and conclude this paper in Section 6.

2. IT sophistication An organization’s information technology sophistication (or IT maturity) depends not only on the technological aspect but also on various organizational characteristics [18]. Identifying the determinants of IT sophistication requires a thorough literature review from various perspectives. Nolan [48,49] used his stage hypothesis model to develop IS maturity construct in the context of IT adoption by organizations. Subsequently, a nine-item instrument was developed [7] for measuring IT maturity to classify firms as bmore matureQ and bless matureQ with references to Nolan’s stage model and other studies. Raymond and Pare [54] define IT sophistication as a multi-dimensional construct, which includes aspects related to technological support, information content, functional support, and IT management practices. In this study, we focus on determining what organizational and technological factors are considered explicitly or im-

791

plicitly critical in today’s business environment for measuring IT sophistication in the context of adopting Web services in three application domains—intranet, extranet, and Internet. 2.1. Organizational factors 2.1.1. Financial and technology resources One of the usual purposes of adopting a new technology is to gain a competitive advantage. There is a body of knowledge in the IS literature that attempts to explain how IT helps gain a competitive advantage using the resource-based view of the firm [5]. According to this view, competing firms vary in the resources they possess, and heterogeneity and immobility of those resources are the sources of competitive advantages. From a review of the extant literature, three resources emerged to be critical in this context—financial, technological, and managerial resources. bAvailability of technical and financial resourcesQ is cited as one of the important factors for the adoption of electronic data interchange (EDI) in small firms [36]. The significance of financial resources in the corporate adoption of Web technology is also stressed [40]. This study suggests a strong correlation of a firm’s Web presence and its financial strengths. Firm size is often considered a proxy of financial and technological resources, and is hence believed to be positively associated with IT adoption. In the research on IS innovation in organizations [60], it was found that the early adoption of IS innovation is more likely to happen when both the organization and its IT department are large. Eder and Igbaria [24] found that firm size is positively correlated with intranet diffusion. The relationship between size and innovation under different moderating factors is further explored in [22]. 2.1.2. Top management support The final decision to adopt a new technology is typically made by top management no matter how much resources an organization possesses. There are a considerable number of studies that underscore the role of top management support as one of the deciding factors for the success of any IT endeavor [24,40,70]. Leadership and strategic direction of management are cited as the most important factors for corporate adoption of Web technology [40]. Two case studies are

792


conducted [70] to compare the adoption of intranet using a modified version of a previous IT implementation model [21]. In that study, top management support emerged as a critical factor in the diffusion phase as well as in the fit stage, which is described [70] as the stage where the innovation is tailored to meet the organization’s specific requirements. 2.1.3. IT management maturity IT management maturity is examined [31] for typifying firms in terms of their evolution in the planning, organization, control, and integration aspects of there IS function. Greater IT management maturity can be characterized by IT managers’ awareness of the firm’s long-term strategic plans, explicit consideration of the firm’s future plans during IS planning, and evaluation of IS performance being based on its contribution to the overall firm’s objectives [13,23,42, 58,66]. In addition, in firms with a high level of IT management maturity, top management may be expected to have greater knowledge about IT and participation in IS planning [38,42,57]. 2.1.4. IT competency of human resources It is suggested [68] that, in order to achieve competitive advantage, organizations will make substantial investments in IT, and maintaining qualified IT professionals of high caliber is critical. This challenge requires managers to consider workers as organizational resources from which competitive advantage can be obtained within a particular market [5]. Addis [4] emphasized that basic skills of workforces are an element of competitiveness for organizations because basic skills encompass personal development, occupational skills, and IT skills. Managers often cite the sophisticated use of IT as the key reason behind the increase in IT skills requirement on workers [11]. Chen et al. [16] also suggested that IT skill set is one of the factors affecting adoption and diffusion of innovations for e-business systems. 2.2. Technological factors Apart from the organizational factors discussed above, technological factors are necessary to estimate the IT sophistication of an organization. Various technological factors are considered important for businesses’ IT adoption and implementation.

2.2.1. Fundamental information technologies Fundamental information technologies such as Web and Internet-related technologies are crucial for successful implementation of Web services applications. A survey was conducted [59] to identify the key information technologies that have the greatest organizational impact. Three groups of technologies emerged as critical from their study: (1) human interface technologies, (2) communication technologies, and (3) systems support technologies. Various hardware and software components that are considered to be critical for businesses include office automation systems, storage and compression devices, networking devices, and decision support systems [30,51,59,]. Since businesses need to transfer huge amounts of data within and across organizations, bandwidth is another important determinant [12,27]. Content management has also been identified as an important aspect of companies’ Web-based applications. Basically, content management helps companies to use a logical, consistent, and effective view to publish content on their Web applications. When e-business crosses enterprise boundary for extranet and Internet applications as Web services grow in the future, content management will become even more critical [25,33,45]. Moreover, the security of information to be transferred and stored is of utmost importance [47,55]. The use of technologies like firewall, SSL (Secure Socket Layer), encryption, and digital signature can be considered primary indicators of the security measures the organization undertakes to protect its data. 2.2.2. Web services specific infrastructure and tools In the Web services context, related infrastructure such as EAI and SOA (Services-Oriented Architecture) can effectively facilitate Web services development and implementation [1,10,67]. Specific technologies and standards such as SOAP, WSDL, UDDI, and XML are natural choices to adopt Web services applications [41,63]. New Web services-related tools and platforms should also be considered in order to facilitate effective Web services implementation. Tools (e.g., Microsoft.Net Studio, Sun One, IBM WebSphere and Web Services Toolkit, Oracle9iAS Web Services Application Server, and BEA WebLogic) and platforms (e.g., .Net and J2EE) can be employed. Chen [15] described the set of standards and supporting technol-


ogies of Web services as the enablers for implementing dynamic e-businesses. 2.2.3. IT readiness and utilization Measurement of IT readiness (also termed as ereadiness or innovation readiness) provides a snapshot of the current utilization of information technology in an organization. While most literature focuses IT readiness at macro level (e.g., country level), we describe IT readiness in the scope of an organization as the degree to which an organization is technologically prepared to participate in the networked environment that demands the adoption of innovations. Jutla et al. [39] define e-readiness as the level of preparedness pertaining to the ability to exploit Internet technology for economic purposes through the rapid adoption of e-business. E-readiness is described as an indicator of the ability to prosper and grow in the Internet economy in [53]. The importance of organizational readiness for successful adoption of IT innovation is indicated in [35]. It is further postulated that a higher level of readiness leads to a lower level of innovation risk and a more successful IT innovation outcome. 2.2.4. IT integration and interoperability To facilitate successful introduction of a new technology into an organization, this new technology has to be compatible with the current business processes, practices, and existing IS applications. This requirement is emphasized in [19,43]. While the former research focused on the adoption of Internet, the work of the latter focused on the importance of different user perceptions in a Group Support System (GSS) setting. The components of an information integration technology platform to provide a robust and unified solution to information integration that can bring new insights and new business opportunities are presented in [56]. A study conducted by the USA National Academy of Engineering [65] identified five primary motivations for further investments in and investigations of systems integrations, and implied that systems integration and interoperability can facilitate adoption of new technologies. For example, three of the five primary motivations include bpropagation of information technologies throughout nations and organizations produces the need for inter-operability and connectivity across equipment and applicationsQ; bthe installed base of

793

information technology products will grow to accommodate both new technologies and new capabilitiesQ; and badvances in information technology and growing appreciation for what ultimately can be accomplished will necessarily promote organizations and nations to search for both new information technology applications and new sources of competitive advantageQ. Thermistocleous and Irani [62] considered important components for application integration adoption. In their proposed model for evaluation and adoption of Enterprise Application Integration (EAI) technology, they cited various technological structural factors such as maintainability, flexibility, scalability, and portability as the necessary requirements for integration. 2.3. Special factors for extranet and Internet Historically, Internet and intranet sites have been managed separately [61]. For example, content management is more complicated than simple Internet, extranet, and intranet distinctions because it is a function of the ownership and intended use of the data. We argue that specific managerial and technological arrangements are needed for implementing extranet and Internet applications even though they are similar to intranet applications. With the advent of the Internet, cyberspace has become a parallel business outlet for most businesses. In this context, the presence of B2B and B2C electronic commerce platforms seems to be key indicators of technology sophistication. A metric is developed [69] to measure the e-commerce capability of organizations that conduct at least part of their businesses over the Internet. They measured such capability metrics along four dimensions: (1) information, (2) transaction, (3) customization, and (4) supplier connection. While a secure Internet site serves as the interface for B2C transactions, Virtual Private Network (VPN) is essential for inter-organizational transactions via an extranet. Moreover, B2B applications often require interoperability of disparate information systems and the presence of Web services offers a unique solution to this interoperability problem by providing an infrastructure for communication of public processes while maintaining the confidentiality of private processes. However, it was found [6] that the adoption of Web technology and electronic commerce practices is still quite primitive in many

794


industries, including those where early adopters have made significant advances.

3. Model To locate the position of a company in the technology adoption space proposed by Chen et al. [17], it is necessary to specify the factors that contribute to each of the three domains identified in their model— intranet, extranet, and Internet. Table 1 summarizes them with a concise list of the various aspects used to measure the blevel of sophisticationQ of a company’s IT applications. We use factors generally adopted by past research discussed in the previous section, and our literature review suggests that there are certain degrees of overlap among the various factors captured by different researchers. Adjusting one factor aiming to improve for a specific domain might affect another factor in the same or other domain. For demonstration purpose, in our model and simulation we derive our results based on the assumption that factors contributing to IT sophistication are independent of each other. Additional details (e.g., some degree of interactively fractional improvement of one factor due to the improvement of another correlated factor) can be built into model and it belongs to our ongoing research. We have grouped the factors into two broad categories—organizational and technological. Some factors are common to all three domains, while the others would be specific to a particular domain. Common factors of these two categories are considered to be the deciding elements in the intranet axis since we assume that the intranet is the most fundamental form of

technology among the three dimensions. For example, IEEE Computer Society defines extranet as ba set of intranets connected for specific objectivesQ and implicitly suggests their complexity differences [37]. In this study, we use the basic distinction of the three domains by referring to their scope (i.e., intranet applications are internally within an organization; extranet applications are externally among organizations; and Internet applications are among any connected parties). It is argued that, in general, complexity of configurations and implementations is increasing from intranet to extranet, and to Internet applications. However, it should be noted that specific cases of intranet applications in practice might be more sophisticated than general extranet or Internet applications in selected functionalities. In this study, we assume that sophistication in either of the other two dimensions requires some additional competence over the intranet capabilities. Therefore, the presence of intranet and extranet applications indicates the organization’s capability of extending its reach into the extranet and Internet dimensions, respectively. Other factors identified from our discussion in the previous section are categorized to be specific to extranet and Internet. To determine the position along an axis, we need to determine the relative importance of each of the factors that constitute the domain as well as the extent to which they are presently diffused. For example, if btop management supportQ considered to be one of contributing factors along the intranet dimension, we need to determine how important it is for Web services adoption in the intranet domain. We also need to find out how much of the needed support actually exists now in the organization. Formally, the Cartesian

Table 1 Factors for measuring the bIT sophisticationQ of a company Common factors for intranet, extranet, and Internet Organizational factors

Technological factors

Financial and technology resources Top management support IT management maturity

Fundamental information technologies Web services specific infrastructure and tools IT readiness and utilization

IT competency of human resources

IT integration and interoperability

Special factors for extranet

Special factors for Internet

Intranet-related implementation

Extranet-related implementation

Extranet-specific applications (e.g., VPN, B2B)

Internet-specific applications (e.g., B2C, B2G)


Completely Web services-enabled

(10, 10, 10)

Q (xd, yd, zd)

z

te

rn

et

A

pp

lic at io ns

d

P (xo, yo, zo)

In

Extranet Applications

y

795

x O (0, 0, 0)

Intranet Applications

Fig. 1. Adoption space model.

coordinate D j of a company along any dimension j (= x, y, and z) will be given by: Ij X

Dj ¼

ai Ti

i¼1 Ij X

S

j ¼ x; y; and z

ð1Þ

ai

pany needs to make in order to achieve the desired amount of Web services adoption (see Fig. 1). So the projected costs of moving from P to Q for a company are estimated by calculating the Euclidean distance as suggested in [17]: qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ð2Þ d ¼ ðxd xo Þ2 þ ðyd yo Þ2 þ ðzd zo Þ2 :

i¼1

where I j = total number of factors to be evaluated in dimension j, a i = the weight or the importance of factor i in the dimension, Ti = degree of adoption/ diffusion of factor i in the dimension, and S = scale to which the distance is compared.1 For a company, the distance from its current position P(x o, y o, z o) to its desired Web services-enabled position Q(x d , y d , z d ) in the adoption space represents the magnitude of the associated costs to reach the desired position. That is, once the coordinates of an organization in the adoption space are found and the destination decided, the distance d between point P and destination Q represents the effort that the com1 We use a scale of 0–10 in this study for simplicity. From Eq. (1) for the position along a dimension, we can use any reasonable value for the scaling factor S and the choice of the value 10 should not affect our findings.

It is assumed in the current study that the higher the level of IT sophistication, the lower the cost to adopt Web services technology. This assumption is based on the argument that the closer a company is to the destination it wants to reach, the easier and cheaper it is to achieve such transition. In a study of EDI adoption by seven small firms, it was found [36] that the relatively low computerization level of the operations of small firms makes the integration of sophisticated IT difficult, which in turn necessitates higher expenditures in capital, people, and technology. In examining the assimilation of Web technologies in 525 companies, the relationship between significant cost savings and the level of sophisticated Web technology assimilation was confirmed [14]. Moreover, other findings [9] suggested that, to integrate more sophisticated IT, a company needs to invest more financial resources in its efforts. Therefore, based on

796


these research findings, in the proposed model, we assume a reverse relationship between the level of an organization’s IT sophistication and its adoption costs of Web services technologies. If different costs are incurred for the adoption along the three dimensions, then the distance will be measured by taking into account the different weights for the three dimensions. For example, if the weights are w x , w y, and w z for the x, y, and z axes, the distance would be given by: qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi d ¼ wx ðxd xo Þ2 þ wy ðyd yo Þ2 þ wz ðzd zo Þ2 : ð3Þ Parameters w x , w y, and w z represent the proportion of the total costs required to make one unit change in the x, y, and z axes, respectively (we would use x, y, and z to represent the intranet, extranet, and the Internet domain, respectively). For computational purpose but without loss of generality, we use a scale of 0 to 10 to represent the length of each axis. Hence, if a company is at (10,10,10), it is perceived to be completely Web services-enabled given the organization’s own evaluation of the importance of the various factors. However, if the relative importance of these factors changes in the future, the company has to re-estimate its status with respect to Web services adoption. There may be a concern that organizations can move to a goal where bbackwardQ adjustment of one or more dimensions of the adoption space may occur. We conjecture that organizations will not bbenefitQ or bsave costsQ by requiring less sophisticated status. For example, organizations will not gain benefits by getting rid of Web/Internet technology or by reducing top management support. Therefore, by moving backward on (or reverting) one dimension of the adoption space, it simply would not incur extra costs. That is, organizations maintain the status quo of the specific factor even though the new requirement of IT sophistication of that specific factor (due to the movement to the new goal) is less. Therefore, the formulation for calculating Euclidean distance can be modified as: d¼

qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ðxd minfxo ; xd gÞ2 þ ðyd minfyo ; yd gÞ2 þ ðzd minfzo ; zd gÞ2 ðwithout weightsÞ

ð2aÞ

d¼

qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi wx ðxd minfxo ; xd gÞ2þwy ðyd minfyo ; yd gÞ2 þ wz ðzd minfzo ; zd gÞ2 ðwith weightsÞ:

ð3aÞ

Since the objective of the current study is to provide management with guidelines for formulating effective Web services adoption strategies, one prerequisite is for the organization to evaluate its current position in the adoption space by measuring the level of IT sophistication. While the proposed model does not specifically set out to address this evaluation issue, some discussions are warranted. It is noted that the factors identified in Table 1 to have influence on Web services adoption are heterogeneous in nature. Some of the factors focus on the perceived characteristics of technology and some on organizational attributes, while the others cope exclusively with inter-organizational coordination. Owing to their inherent differences in nature, the factors require different mechanisms for measurement. The instruments that have been employed in previous IT diffusion literature to measure these factors include: (1) the percentage of IT staff with given years of experience with the technology for IT skill set; (2) an ordinal variable representing the level of IT spending for financial resources; (3) the number of underlying necessary technologies acquired for technological resources; (4) a five-point Likert scale to weigh top management support; and (5) the Guttman scale with five or seven levels to represent progressive stages from no awareness to general deployment for IT utilization/ assimilation [20,26,36]. For the other factors, there are survey questions that management can answer so as to evaluate the degree or level of intensity in a specific factor after combining or collapsing their answers. For instance, it has been proposed [14] to use seven questions with a seven-point Likert scale to measure IT sophistication, ranging from operational costs reduction to improved service to customers. Most of the instruments have been validated as effective and reliable measures. Management can employ them to evaluate the diffusion level of each factor for their organization, i.e., Ti for factor i in Eq. (1). It is important to note that evaluating these factors is company-dependent, and management has the flexibility to adjust the measurement scales used, assign varying weights a i to the factors, and decide the ultimate destination S the organization would like to reach in a manner considered appropriate for their


organization. Once the relevant parameters are decided and the factors are measured, management can use the proposed model to find the cost-effective route to Web services adoption. With the assistance of the adoption model, managers can also perform sensitivity analysis dynamically during the course of actions when the circumstances change and require fine-tuning of the originally chosen strategy. 3.1. Research goal Once an organization evaluates its current position in the technology adoption space, the next step is to decide on an effective strategy to achieve the desired level of sophistication. Given the relative significance (weight) of the factors along each dimension, an organization can advance its position by improving the diffusion level of the contributing factors in each dimension. However, this requires both financial and human resources. Working simultaneously on all of the factors may require huge capital investment. It would be more practical to identify the factors that take the company closest to its destination given the limited resources. One of the aims of our research is to help decision makers choose the best strategy for achieving the organizational goal. It must be noted here that, for the long term, it is always recommended that the company enhance all of the factors eventually. Our research involves the short-term scenario where the highest return is sought given the limited amount of resources. The questions we attempt to answer in this study are: (1) With limited resources, what is the best strategy for a company to move from its current position to the desired destination given a company’s chosen weights and current levels of diffusion for the factors along the three dimensions? (2) How do the relatively high/low values of the weights and the relatively high/low level of diffusion along a particular dimension affect the adoption strategy? Since the relative importance of the factors (i.e., the weights) can be assumed to remain the same in the short term, it seems logical that the most likely candidates for improvement are the ones that have the highest importance (i.e., the factors with the highest overall weight).

797

However, choosing factors only on the basis of weights might not result in the desired effect because distances of each of the dimensions from the destination should be considered as well. If one of the dimensions is relatively farther from the destination than the other two, it is also important to consider improving this dimension to reduce the overall distance from the destination. Because our research intends to identify the best strategy under limited resources, we formulate our strategies based on the two most likely choices. While it might be possible to improve more than these two factors if resources were plentiful, the choice of these two factors–the factor that has the highest overall importance and the factor that has the highest importance in the dimension that is farthest from the destination–are the ones that are most relevant to our research question (i.e., with only limited resources). Following this line of thinking, we consider the following three strategies in this study: (1) maxW strategy—an organization focuses all its attention and resources on improving the factor that has the highest overall importance (maximal weight) among all the three dimensions. In other words, the factor having the global highest weight is chosen and all resources are spent on improving the diffusion level of the factor in the dimension it belongs to. We name this factor the maxW factor. (2) minD–maxW strategy—an organization focuses all its resources on improving the factor having the highest weight along the dimension which is comparatively farthest from destination. In other words, an organization identifies the dimension that needs to be improved most and then targets the most important factor on that dimension to improve. For example, if the Internet dimension is the farthest from the destination (compared with intranet and extranet dimensions), we refer the Internet dimension as the minD dimension. The factor that has the highest weight on this Internet dimension is the minD–maxW factor and will be targeted for improvement under the minD–maxW strategy. (3) Distributed strategy—instead of focusing only on one dimension and one factor, an organization allocates available resources to both the maxW factor and minD–maxW factor.

– 0.61 0.36 – 0.73 0.26 0.51 0.72 0.81 0.28 0.85 0.30 0.36 0.88 0.28 0.51 0.74 0.11 0.04 0.68 0.50 0.62 0.76 0.71 0.18 0.63 0.55 0.24 0.85 0.45

Simulation approach is often used when people have complex problems that cannot be solved by other exploratory means such as optimization modeling, survey, and case study. In this study, we utilize the simulation approach mainly for the discovery and formalization purposes [28]. We try to set up random variations that may exist simultaneously in many factors affecting technology adoption in real life. More specifically, with many possible variations in IT status, we would like to identify the factors that would be most effective in helping an organization to achieve its goal of adopting Web services.

Table 2 Weights of factors for a given company

3.2. Research design

Top management support

IT management maturity

IT competency of human resources

Fundamental information technologies

Web services specific infrastructure and tools

IT readiness and utilization

IT integration and interoperability

Intranet- or extranet-related implementation

We use a numerical example to clarify these three strategies. Table 2 illustrates the weights (importance) of the factors along the three dimensions for a given firm where the first four factors represent organizational factors, the next four factors represent technological factors, and that last two special factors for either extranet or Internet. The current position of a company is (5.3, 7.21, 4.14) in the adoption space and the destination is (10, 10, 10); therefore, the Internet dimension (z axis) is the farthest from the destination and thus it is the minD dimension. Moreover, the factor with the highest importance of the Internet dimension (i.e., the eighth factor–IT integration and interoperability–with 0.81) is our minD–maxW factor. On the other hand, the sixth factor (Web services specific infrastructure and tools) of the extranet dimension ( y axis) has the highest overall importance among all the three dimensions (weight = 0.88) and is regarded as the maxW factor. We conjecture that, with limited resources, management needs to decide whether they should concentrate on the maxW factor, focus on the minD–maxW factor, or distribute their resources between the maxW factor and the minD–maxW factor. The three strategies described above provide a comprehensive set of alternative choices for resource allocation in the context of Web services adoption. Given the current position, destination, and the resources to reach the destination, an organization’s aim will be to get as close to the goal as possible. However, since resources are limited, improving the factors with high significance would have more impact. Hence, management has to decide on the tradeoff between these two likely candidate factors.

Intranet Extranet Internet

Extranet- or internet-specific applications


Financial and technology resources

798


As discussed in the previous section, we are interested in investigating the relative effectiveness of the three strategies under varying combinations of weights and diffusions. In this study, the improvements are made at the individual factor level but not at the category level as the parameters are not treated as interdependent (i.e., the weights and diffusion levels of the factors in one dimension do not affect those in the other two dimensions). We restrict the combinations of weight and diffusion values of factors on one specific dimension in order to get meaningful results. However, since the space model is symmetric, the findings will still hold if we switch restriction from one dimension to another. We do not choose the Intranet dimension in this study because it has a smaller number of factors (i.e., eight factors). We choose extranet, which has 10 factors, and we could have chosen Internet, which has 10 factors as well. The values of the weights and diffusion levels along each dimension are generated randomly from a uniform distribution. The data generation and analysis are performed using Visual Basic and Microsoft Excel. We draw the values of weights and diffusions for each factor from 0 to 1 and they are randomly generated from a uniform probability distribution. We investigate the following three different scenarios with three different combinations of weights and diffusions along the extranet dimension: (1) HW-HD scenario (high weight and high diffusion on one dimension, any weights on the other two dimensions): the first scenario accounts for the case where the factors determining IT sophistication in this dimension are perceived important by management, and the attention paid before to those factors has resulted in high diffusion. (2) LW-LD scenario (low weight and low diffusion on one dimension, any weights for the other two dimensions): this scenario indicates that there are factors that have been overlooked by an organization and hence have a low level of diffusion. (3) HW-LD scenario (high weight and low diffusion on one dimension, any weights on the other two dimensions): this scenario, on the other hand, represents the situation where the significant factors have been identified, but the level of expertise or presence in these areas is cur-

799

rently low. This may happen if the critical technologies or applications are relatively new to the organization. We do not consider the fourth possible scenario where the factors have a low significance level but high diffusion for it is a highly unlikely situation. In reality, if the organization does not consider the technologies to be important, then there is less chance that it would spend resources on those technologies, which already have a high diffusion level. In the simulations, we restrict high values in the range (0.6, 0.9) and the low values in the range (0, 0.3). As mentioned before, we make these combinations along the extranet dimension. For our investigation purpose, we specify the final destination as (10, 10, 10). This is chosen only for computational demonstration since our aim is to investigate the effects of weights and diffusions of various factors on adoption costs, which are proportional to the distance between the current position and a given destination. This assumption thus would not alter our findings had we used a different destination. On the other hand, our simulation program is designed to have the flexibility to use any destination point other than (10, 10, 10). Sometimes, it might be necessary to have different destinations to reflect different goals of companies. For example, when a company wants to concentrate more on adopting Web services technology for intranet applications but not so much for extranet or Internet applications, this company can set its destination at, say, (10, 5, 5) instead of (10, 10, 10). For each scenario (HW-HD, HW-LD, LW-LD), we simulate 500 runs and implement all the three strategies—the maxW, minD–maxW, and distributed strategies. In each run, the distance improvement by each individual strategy relative to the destination is calculated. The strategy that takes the company closer to the destination (i.e., desired Web services adopTable 3 Strategy description: main experiment Resources dedicated to

maxW strategy minD–maxW strategy Distributed strategy

maxW factor

minD–maxW factor

2 0 1

0 2 1

800


Table 4.1 Frequency counts of simulation results for strategies under different scenarios Diffusion

Weight

High

Low

High

Low

maxW strategy is best: 38 minD–maxW strategy is best: 189 Distributed strategy is best: 16 maxW strategy = minD–maxW strategy: 2 All three strategies are equal: 255 Best strategy = minD–maxW N/A

maxW strategy is best: 229 minD–maxW strategy is best 176 Distributed strategy is best: 30

All three strategies are equal: 65 Best strategy = maxW maxW strategy is best: 29 minD–maxW strategy is best: 461 Distributed strategy is best: 10 Best strategy = minD–maxW

tion) is considered to be the best choice. Since we are seeking the best strategy for the company in the short run, the weights of factors are kept constant,

and we vary the levels of diffusion to find the best strategy. The argument to keep weights fixed is that in the short term the relative importance of various factors tends to remain unchanged. Management can take measures to enhance the diffusion levels of the most important factors to improve an organization’s IT sophistication for Web services adoption. Also, another practical aspect of the adoption problem is limited resources, so we allow only a fixed amount of resources for all the scenarios. However, we also perform a sensitivity analysis to ensure that the findings are not dependent on the amount of resources allowed. That is, we first conduct our main experiment allowing only 2 units of resources, and then we perform a sensitivity analysis allowing 3 units of resources. The results of our simulation and their implications are provided in the next section.

4. Simulation results 4.1. Main experiment In the main experiment, the available resources are 2 units. Pertaining to the strategies we discussed in Section 3.1, the resource distribution for the three strategies is as given in Table 3.

Table 4.2 Statistics of weights and diffusions for three scenarios Mean weight

Std. deviation of weight

Mean diffusion

Std. deviation of diffusion

Panel A: high-weight/high-diffusion scenario maxW strategy is best 0.584079 minD–maxW strategy is best 0.573505 Distributed strategy is best 0.593393 t-statistica 0.224395

0.265245 0.251979 0.259537

0.577754 0.577698 0.577857 0.002055

0.259830 0.264474 0.261646

Panel B: high-weight/low-diffusion scenario maxW strategy is best 0.575711 minD–maxW strategy is best 0.613466 Distributed strategy is best 0.574893 t-statistica 1.44012

0.273461 0.251979 0.262742

0.359599 0.384523 0.362452 0.463547

0.280535 0.293208 0.284814

Panel C: low-weight/low-diffusion scenario maxW strategy is best 0.365308 minD–maxW strategy is best 0.373528 Distributed strategy is best 0.354250 t-statistica 0.149340

0.287309 0.290613 0.285184

0.385037 0.369116 0.352357 0.154561

0.289467 0.289953 0.267763

a

t-test for paired two samples for mean.

A.N.K. Chen et al. / Decision Support Systems 42 (2006) 789–809 Table 4.3 Mean distance and differences from the best strategy for three scenarios Mean distance Mean difference Std. deviation from destination from the best of the mean strategy difference Panel A: high-weight/high-diffusion scenario maxW strategy 7.561531 0.052522 minD–maxW 7.513213 0.004205 strategy Distributed 7.534985 0.025976 strategy 12.508300*** t-statistica Panel B: high-weight/low-diffusion scenario maxW strategy 11.077040 0.013419 minD–maxW 11.089450 0.025838 strategy Distributed 11.080930 0.017314 strategy t-statistica 5.822084*** Panel C: low-weight/low-diffusion scenario maxW strategy 11.117752 0.098484 minD–maxW 11.021620 0.002351 strategy Distributed 11.066200 0.046932 strategy t-statistica 34.588740***

0.084472 0.018032 0.040390

0.023104 0.040805

Table 5 Strategy description: sensitivity analysis Resources assigned to

maxW strategy minD–maxW strategy Distributed strategy A Distributed strategy B

maxW factor

minD–maxW factor

3 0 1 2

0 3 2 1

result. Since such cases suggest no differences in choosing either one of the other two strategies, we exclude them from further analysis. After excluding the cases where two or more strategies yield the same best results, we further investigate whether the resulting distance difference between the

0.018760

0.061011 0.011831 0.029324

Table 6.1 Frequency counts of simulation results for strategies under different scenarios Diffusion

Weight

High

a

t-test for paired two samples for mean. *** Significant at p b 0.01 level.

Table 4.1 provides a summary of results for the main experiment.2 The relative frequency distributions of the strategies to move the organization closer to the destination suggest that, for both HW-HD and LW-LD scenarios, it is advisable to spend the available resources on the dimension that needs the most improvement. However, the findings are different in the HW-LD scenario, where focusing on the factor with the highest overall weight proves to be the best strategy. Interestingly, the distributed strategy has the fewest frequency counts in each of the scenarios. Table 4.1 also provides the counts for the cases where two or more strategies provide the same best 2

801

It is noted that these strategies sometimes generate the same best result when the random numbers are generated in a way that the maxW factor happens to be the minD–maxW factor as well; i.e., the factor with the highest overall weight falls in the dimension that is farthest from the destination. This is confirmed by checking the simulation results programmatically. These results are excluded from further analysis since they are of no practical use for our study.

Low

High

Low

maxW strategy is best: 24 minD–maxW strategy is best: 137 Distributed strategy A is best: 22 Distributed strategy B is best: 12 MaxW strategy = minD–maxW strategy = distributed strategy A: 7 MaxW strategy = minD–maxW strategy = distributed strategy B: 4 All four strategies are equal: 294 Best strategy = minD–maxW N/A

maxW strategy is best: 206 minD–maxW strategy is best: 151 Distributed strategy A is best: 30 Distributed strategy B is best: 31 maxW strategy = minD–maxW strategy = distributed strategy A: 1

All four strategies are equal: 81 Best strategy = maxW maxW strategy is best: 33 minD–maxW strategy is best: 438 Distributed strategy A is best: 6 Distributed strategy B is best: 23 Best strategy = minD–maxW

802


strategies in a particular scenario is caused by the difference in the initial weights/diffusions or by the strategies themselves. We present the statistics of weights and diffusions in the three scenarios of the simulation in panels A, B, and C of Table 4.2. The mean weight and mean diffusion as well as their standard deviations are provided for each scenario and for each strategy. A preliminary observation suggests that the weights and diffusion between the strategies are not significantly different. We conduct t-tests for paired two samples for the means for statistical inference. The paired t-tests are performed on the two strategies having the highest and second highest frequency counts in each scenario. Since in all the cases either the minD–maxW or the maxW strategy proved to be best, the t-tests are performed between these two strategies only, as highlighted in Table 4.2. The results of the t-tests show that there are no significant differences at the p b 0.05 level between the weights and diffusions of the two strategies in any of the scenarios under investigation. This suggests a very low possibility that the distance difference between the top two strategies in a particular scenario is due to the difference in the initial weights or diffusions.

To ascertain the validity of our results, we computed the mean distances from destination for each of the strategies. Panels A, B, and C of Table 4.3 provide the mean distances from the destination (10, 10, 10) after allocating resources according to the three strategies for each of the three scenarios. Supporting the results presented in Table 4.1, these results show that the minD–maxW strategy (spending all the resources in increasing the level of diffusion for the most significant factor along the dimension where the company needs to make the greatest improvement) is a better option for both the HW-HD and LW-LD scenarios. The maxW strategy (spending all the resources in increasing the level of diffusion for the factor with the highest overall weight) emerges as the dominant option for the HWLD case. Further, Table 4.3 also presents mean differences from the best strategy as well as their standard deviations for each of the three scenarios. The mean difference from the best strategy is calculated as follows. For each simulation run, the difference of each strategy from the best strategy is measured. Table 4.3 presents the average value of these differences from 500 runs for each strategy. The results show that the difference of mean differences of maxW and minD–

Table 6.2 Statistics of weights and diffusions for three scenarios Mean weight

Std. deviation of weight

Mean diffusion

Std. deviation of diffusion

Panel A: high-weight/high-diffusion scenario maxW strategy is best 0.584256 minD–maxW strategy is best 0.574299 Distributed strategy A is best 0.573003 Distributed strategy B is best 0.592173 t-statistica 0.173026

0.259180 0.265010 0.268405 0.269978

0.575283 0.573115 0.574481 0.596786 0.019082

0.263620 0.263692 0.269611 0.271820

Panel B: high-weight/low-diffusion scenario maxW strategy is best 0.574714 minD–maxW strategy is best 0.614818 Distributed strategy A is best 0.592048 Distributed strategy B is best 0.569827 t-statistica 1.410230

0.277089 0.256586 0.268930 0.272645

0.353240 0.388806 0.368298 0.371797 0.614348

0.278795 0.301719 0.290379 0.289489

Panel C: low-weight/low-diffusion scenario maxW strategy is best 0.338994 minD–maxW strategy is best 0.377454 Distributed strategy A is best 0.339048 Distributed strategy B is best 0.354705 t-statistica 0.789061

0.267812 0.297668 0.284415 0.280596

0.362576 0.369211 0.381429 0.394876 0.068628

0.277903 0.289088 0.286195 0.284817

a

t-test for paired two samples for mean.


maxW strategies are statistically significant at the p b 0.01 level for all three scenarios. 4.2. Sensitivity analysis We next perform a sensitivity analysis of our experiment to ascertain that our findings are not a function of the amount of available resources. That is, we make sure that consistent results are obtained when more resources are available. For this sensitivity analysis, we increase the available resources to 3 units. The possible resource allocations for the strategies are given in Table 5. Note that there are now two distributed strategies as two combinations of the available

Table 6.3 Mean distance and differences from the best strategy for three scenarios Mean distance Mean difference Std. deviation from destination from the best of the mean strategy difference Panel A: high-weight/high-diffusion scenario maxW strategy 7.389655 0.058640 minD–maxW 7.335272 0.004257 strategy Distributed 7.349542 0.018527 strategy A Distributed 7.367692 0.036677 strategy B t-statistica 10.571136*** Panel B: high-weight/low-diffusion scenario maxW strategy 11.036760 0.020435 minD–maxW 11.053610 0.037288 strategy Distributed 11.043600 0.027273 strategy A Distributed 11.037970 0.021651 strategy B t-statistica 5.644436*** Panel C: low-weight/low-diffusion scenario maxW strategy 10.990860 0.145696 minD–maxW 10.851240 0.006078 strategy Distributed 10.890620 0.045458 strategy A Distributed 10.937190 0.092026 strategy B t-statistica 31.411573*** a t-test for paired two samples for mean. *** Significant at p b 0.01 level.

0.113778 0.0196953 0.034998 0.036677

0.035593 0.056485 0.033586

803

resources are possible compared with only one distributed strategy in the main experiment. The parameter variations in the sensitivity analysis are similar to the main experiment. In all three scenarios, 500 simulation runs are also performed. The frequency counts of each of the strategies in the three scenarios are shown in Table 6.1. These results support our previous findings. Consistent with the results of our main experiment, the minD–maxW strategy is the best strategy most frequently for both the HW-HD and LW-LD scenarios. The best strategy for the HW-LD is again most likely to be the maxW strategy and, in none of the scenarios, the distributed strategies emerge as the most frequent best one. We perform similar t-tests on the weight and diffusion values to ascertain that differences in their initial values do not cause one strategy to dominate the others. The results are presented in panels A, B, and C of Table 6.2. We compare the differences of the parameters for the best and the second best strategies. In all the scenarios, the differences between the mean values of weights and diffusions are non-significant at p b 0.05. Finally, we perform the distance analysis resulting from the implementation of the strategies. Panels A, B, and C of Table 6.3 present the data from the distance analysis. For each scenario, the mean distance from the destination after each strategy implementation is calculated, as well as the average distance from the best strategy. The mean differences between the best and the second best strategies are all significantly different at the p b 0.01.

5. Discussions and analyses

0.023234

0.096609 0.023341 0.029902 0.062170

Taken together, our results suggest that, in both the HW-HD and the LW-LD scenarios, it is better to focus on the dimension that needs the most improvement, while in the HW-LD scenario the preferred strategy is to spend resources on the factor that has the highest overall importance. Here we discuss the practical significance of our findings. 5.1. High-weight/high-diffusion scenario The HW-HD scenario represents the case where the factors along the test dimension have great signifi-

804


cance and the level of expertise in these factors is already quite high. Having both high significance and high levels of penetrations makes this test dimension more likely in the first place to be much closer to the destination than the other two dimensions. Thus, spending the resources on the other dimension that needs the most improvement (i.e., the minD dimension) very likely has a greater overall impact on moving the company closer to the destination in the adoption space. The common business wisdom emerging from this scenario is that, if the factors determining IT sophistication along one dimension are very significant and high levels of strength already exist in those areas, attention must be directed to the other weaker dimensions. This HW-HD scenario is elaborated with two examples. The first example considers the case of a leading organization specializing mostly in B2B e-commerce and providing services to other businesses. For such a company, extranet-related factors (e.g., VPN, B2B applications) are considered most important (i.e., of the highest weight) as per the IT sophistication framework. As a major player, the company is also likely to have significant expertise (i.e., high diffusion) in these areas. In order to adopt Web services, the organization should then concentrate on its weakest area with respect to Web services—most likely the Internet. Spending the available resources on the most important factor in this dimension (e.g., providing proper interface and publishing its services) will be the smartest move for the company. SABRE fits into this scenario [64]. The company is a premier provider of travel services to various businesses and travel agencies. To utilize the advantages of Web services, SABRE replaced its partner-specific custom application programming interfaces with Web services and thus was able to create stronger ties with partners and B2B customers by simplifying the technical demands placed on working with and buying from the company. A second example for this HW-HD scenario is an organization that has high technological capability in certain business areas (e.g., financial systems) and attempts to make the best of their expertise by exposing them as Web services. The IT sophistication framework implies that the most important factors for this particular organization are the technological factors (e.g., fundamental technologies, interoperability, etc.) along the Intranet. In-house expertise implies

that high levels of diffusion exist for these factors. In this case, the relatively weaker dimension is also the Internet along which capabilities are required to publish their services. Citibank’s CitiConnect [34] exemplifies this scenario. The company leveraged its skills in electronic payments by introducing the XML-based payment processing service that can be used by existing trading applications. In summary, the HW-HD scenario typically represents an organization that has some sort of technologically advanced solutions and expertise in areas of its core competence. Consequently, for higher short-term returns, the company should concentrate on other relatively weaker areas and choose the most important one for improvement. 5.2. Low-weight/low-diffusion scenario Similar explanations can be provided for the LWLD scenario’s best strategy being the minD–maxW strategy. However, one difference in this case is that the test dimension, having both low weight and low diffusion, is likely to be the farthest from the destination. In other words, the test dimension is most likely the minD dimension itself. Because of the great distance of this test dimension from the destination, devoting all available resources to this dimension produces the greatest overall effect. Therefore, attention must be focused on this weak area that is lagging most for the organization. To illustrate, let us consider a manufacturer connecting its customers via a portal and providing online order-tracking system [3]. The IT sophistication framework suggests that the significant factors for consideration in this case are the B2B applications along the Internet dimension. Nonetheless, if acceptable levels of service already exist, occasional problems may arise from unexpected demand surges. As the traffic to the direct ordering system increases, the inventory management system operating in the backend should be able to handle the increased amount of information and provide updated status synchronously with the ordering system. The efficiency of the inventory management system along the intranet would not seem a significant factor in this particular context and would result in low levels of diffusion. The ensuing weakness, however, creates trouble in the overall performance when adequate levels of expertise exist in other seemingly more important


dimensions. Along the same lines, a second factor that might seem understated is IT management’s visibility to organizational systems and their interrelations. Lack of visibility is cited as one of the major pitfalls in the implementation of Web services [32,50]. The straightforward nature of Web services, incremental approach, and low cost requirement also make the adoption process likely ad hoc and opportunistic, leading to a fragmented state of Web services within an organization. Discounting the role of management visibility will cause problems in the long run. In other words, the LW-LD scenario signifies the cases where expertise along a certain dimension does not seem important at first glance. Low levels of expertise in these areas have a ripple effect on the performance; i.e., the shortcoming(s) of one system affects other systems linked with it. Consequently, this creates a dragging effect—impeding the overall efficiency of operation because of weak spots. It must also be noted that the low-weight low-diffusion scenario is relevant only when the other more important dimensions already have sufficient skill levels and, as such, are easy to be overlooked. This scenario reflects the indirect effects of having lower penetration in certain areas, rather than the direct effect of visibly important factors. 5.3. High-weight/low-diffusion scenario On the other hand, the HW-LD scenario yields different results. In this scenario, the best strategy is to devote all the available resources to the factor having the highest overall significance (i.e., the

805

maxW factor). The HW-LD scenario represents the case where the significance of the factors in the test dimension is high, but there is no sufficient level of penetration in those areas. Owing to the opposite influences of high weights and low diffusion levels on deciding the coordinate of the company in the test dimension, it is difficult to foretell which one of the three dimensions is likely to be the minD dimension. As such, the HW-LD scenario presents a more ambiguous case, and our simulation results suggest that under this circumstance, pursuing the clear-cut maxW strategy to focus on the factor with the highest overall significance would more likely lead to the best outcome. In other words, spending all the available resources on the maxW factor will take the organization much closer to the destination than other options. For this case, it would be advisable to concentrate on the factor with the highest overall weight. For example, consider an organization willing to improve its decision making abilities by the integration and analysis of data and information gathered from its various business applications, transactions, and information systems. The IT sophistication framework would suggest application integration along the intranet dimension to be the most important factor. Nonetheless, since technology-enabled decision making was not in the organizational agenda previously, its diffusion levels along this dimension would be low. Under such circumstances, it is imperative that the company takes immediate care of the most important areas first. British American Tobacco’s Web service adoption illustrates this situation [52]. As the company attempted to enhance its supply chain performance

Table 7 Summary of practical implications of the results Scenario

Likely situation

Strategy

Real-world cases

High weight–high diffusion

An organization wants to expose some of its core capabilities as web services An organization has standard skill levels in most of the areas, some of the areas seem to be not so significant, and hence has a low level of expertise An organization has recently identified some new areas, where rapid diffusion of skill levels are necessary

Allocate all available resources in the weakest areas

SABRE CitiConnect

If all other areas are comparatively stronger, the low expertise areas will create an overall dragging effect. Visibility into the infrastructure is important The areas recently identified are highest priority. All resources should be spent in the highest priority areas, irrespective of the status in other areas

Manufacturing company providing on-line order tracking system to customers via a portal

Low weight–low diffusion

High weight–low diffusion

British American Tobacco Avnet

806


by analyzing data stored in different vendor-specific applications, integrating the data via Web services became the first priority as well as the major stepping stone towards achieving their goal. A similar example in the B2B context can be found in Avnet [8]. Avnet, one of the world’s largest distributors of electronic components and computer products for industry leading manufacturers, needed to create an e-business portal that allows customers and suppliers to configure and place orders, retrieve realtime pricing and availability information for products, and view their order status. To provide customers and suppliers with a broad range of online services, Avnet needed to connect the portal to various backend systems using a variety of protocols and file formats. In this case, the integration of information over the extranet became the most important issue. To sum up, the high-weight/low-diffusion scenario represents the case where specific areas are recently identified to be most important in an organization’s IT infrastructure arising from new business or technology needs. Consequently, the level of expertise for these factors is lacking. Hence, these factors become the first priority for the organization to act upon and spending all available resources on the most important factor provides higher returns. The findings from our simulation results have important implications for helping businesses decide the best strategy to effectively allocate resources for Web services adoption. Table 7 summarizes these practical implications. When an organization decides to invest in improving its current status of IT with respect to adopting Web services but it does not have sufficient resources to invest in all weak areas, the most cost efficient strategy to allocate the resources is a function of the relative weights and diffusion levels of various factors that are critical to the successful adoption of the technology. Different combinations of the weights and diffusions would require different strategies. Allocating resources on an ad hoc basis to the weakest factors may not always lead to the best utilization of available resources. In our simulation, we make the assumption about resource constraint that only two factors can be improved at the same time. Although it is possible that the available resources can be deployed in more than two areas, we argue that the abundance of resources makes the allocation problem less difficult. In that

sense, our results shed light on the decision making process when the organization is faced with a scarcity of resources to adopt Web services. For simplicity, our simulation evaluated only changes of up to 2 or 3 units. Improving the diffusion level in any area requires resources. If the organization has enough resources, over-improving some factors while neglecting others might create imbalance among the key IT areas. Our ongoing research will extend our simulation model to study how adjusting each individual factor will affect a company’s effort to achieve its goal of adopting Web services-enabled technologies. We can further relax the assumption that a company can make improvements on only two factors simultaneously. Future research directed towards investigating the optimal allocation of resources to improve multiple factors simultaneously will have more flexibility to set technology priority. Optimization models of resource allocation using mathematical programming or goal programming for further analyses can be designed. Since we argue that the distance traveled by an organization is proportional to the cost associated with the adoption decision, our research can be extended to estimate the potential expenses. Another complementary research area is to measure the benefits of Web services adoption so that it would enable businesses to conduct a cost-benefit analysis of such adoption decisions.

6. Conclusion In this paper, we first attempt to formalize a way to locate a company in a Web services adoption space based on a company’s bsophistication levelsQ of its current IT infrastructure, factors, and applications. We identify the determinants of IT sophistication along three dimensions—intranet, extranet, and Internet. Further, our research aims to help businesses in the adoption decision of Web services. Guidelines are provided to help businesses pursue the adoption most effectively and choose specific factors for further improvement under the constraint of limited resources. Our results suggest that different strategies should be employed while organizations consider their existing organizational IT status and focused area. First, when the factors determining IT sophistication are already apparent to the management and


attention has been paid to those areas with a resultant high diffusion, investing all available resources to improve the factor with the highest weight in the identified dimension which needs to be improved most would better position a company to adopt Web services. By the same token, in the scenario where there may be technologies that are less important for IT activities of an organization and have a low level of diffusion, investing all available resources to improve the factor with the highest weight in the identified dimension which needs to be improved most would also better position a company to adopt Web services. Finally, when significant factors have been identified but the levels of diffusion in these areas are low (e.g., the technologies are relatively new to the organization), investing available resources to improve the most important one among all factors would better position a company to adopt Web services. Acknowledgements We would like to thank three anonymous reviewers for their critical and valuable comments. These comments have significantly strengthened the quality and presentation of this paper. In particular, we want to express our gratitude to Dr. James Marsden for his insightful suggestions for our initial preparation of this paper. References [1] Accenture, Web services and enterprise integration: friends not foes, http://www.accenture.com/xdoc/en/services/Web/ insights/Web_enterprise.pdf, 2003. [2] Accenture, Web services: IT efficiency today. . .powerful business solutions tomorrow, http://www.accenture.com/xdoc/en/ services/Web/insights/Web_survey.pdf, 2003. [3] Actional, Breaking through the complexity: a project-based approach to realizing the benefits of your Web services implementation, http://itresearch.forbes.com/detail/RES/1105712008_907. html, 2004. [4] M. Addis, Basic skills and small business competitiveness: some conceptual considerations, Education and Training 45 (3) (2003) 152 – 161. [5] J.B. Barney, Firm resources and sustained competitive advantage, Journal of Management 17 (1) (1991) 99 – 120. [6] A. Basu, S. Muylle, Online support for commerce processes by Web retailers, Decision Support Systems 34 (4) (2003) 379 – 395.

807

[7] I. Benbasat, A.S. Dexter, R.W. Mantha, Impact of organizational maturity of information skill needs, MIS Quarterly 4 (1) (1980) 21 – 34. [8] M. Blank, Knowing when to use Web services, Web Services Journal 3 (10) (2003) 26 – 28. [9] L. Bouchard, Decision criteria in the adoption of EDI, Proceedings of the 13th International Conference on Information Systems, 1993, pp. 365 – 376. [10] B. Box, The Web services revolution, http://www.eds.com/ thought/thought_leadership_Web_revolution.pdf, 2003. [11] T.F. Bresnahan, E. Brynjolfsson, L.M. Hitt, Information technology, workplace organization, and the demand for skilled labor: firm-level evidence, Quarterly Journal of Economics 117 (2002) 339 – 376. [12] H. Carr, C.A. Snyder, Management of Telecommunications, Irwin, IL, 1997. [13] J.I. Cash, F.W. McFarlan, J.L. McKenney, L.M. Applegate, Corporate Information Systems Management: Text and Cases, 3rd ed, Irwin, Homewood, IL, 1992. [14] D. Chatterjee, R. Grewal, V. Sambamurthy, Shaping up for eCommerce: institutional enablers of the organizational assimilation of web technologies, MIS Quarterly 26 (2) (2002) 65 – 89. [15] M. Chen, Factors affecting the adoption and diffusion of XML and Web services standards for e-business systems, International Journal of Human–Computer Studies 58 (2003) 259 – 279. [16] M. Chen, A. Chen, B. Shao, The implications and impacts of Web services to electronic commerce research and practices, Journal of Electronic Commerce Research 4 (4) (2003) 128 – 139. [17] A. Chen, R. Labrie, B. Shao, An XML adoption framework for electronic business, Journal of Electronic Commerce Research 4 (1) (2003) 1 – 14. [18] P.H. Cheney, G.W. Dickson, Organizational characteristics and information systems: an exploratory investigation, Academy of Management Journal 25 (1) (1982) 170 – 182. [19] W.W. Chin, A. Gopal, Adoption intention in GSS, relative importance of beliefs, The Data Base for Advances in Information Systems 26 (2&3) (1995) 42 – 64. [20] P. Chwelos, I. Benbasat, A.S. Dexter, Research report: empirical test of an EDI adoption model, Information Systems Research 12 (3) (2001) 304 – 321. [21] R.B. Cooper, R.W. Zmud, Information technology implementation research: a technological diffusion approach, Management Science 36 (2) (1990) 123 – 139. [22] F. Damanpour, Organizational size and innovation, Organization Studies 13 (3) (1992) 375 – 402. [23] M.J. Earl, Experiences in strategic information systems planning, MIS Quarterly 17 (1) (1993) 1 – 24. [24] L.B. Eder, M. Igbaria, Determinants of intranet diffusion and infusion, Omega 29 (2001) 233 – 242. [25] D. Fensel, Challenges in content management for B2B electronic commerce, Proceedings of Second International Workshop on User Interfaces to Data Intensive Systems, 2001, pp. 2 – 4. [26] R.G. Fichman, The role of aggregation in the measurement of IT-related organizational innovation, MIS Quarterly 25 (4) (2001) 427 – 455.

808


[27] J. Fitzgerald, A. Dennis, Business Data Communications and Networking, 7th ed., Wiley, 2001. [28] N. Gilbert, K.G. Troitzsch, Simulation for the Social Scientist, Open University Press, 1999. [29] D. Gisolfi, Web services architect Part 1: an introduction to dynamic e-business, http://www-106.ibm.com/developerworks/ Webservices/library/ws-arc1/, 2001. [30] V. Grover, M. Goslar, Information technologies for the 1990’s: an executive’s view, Communications of the ACM 36 (3) (1993) 17 – 103. [31] Y.P. Gupta, J. Karimi, T.M. Somers, Alignment of a firm’s competitive strategy and information technology management sophistication: the missing link, IEEE Transactions on Engineering Management 44 (4) (1997) 399 – 413. [32] W. Guild, Using Web services to tame technology, http://hbswk. hbs.edu/pubitem.jhtml?id=3195 and t=ecommerce, 2002. [33] J. Hackos, Content Management for Dynamic Web Delivery, John Wiley & Sons, 2001. [34] J. Hagel, J.S. Brown, Your next IT strategy, Harvard Business Review 79 (9) (2001) 105 – 113. [35] B.A. Happ, Applying technology assessment to information systems, in: M.E. Miller, C.A. Romano, B.R. Heller (Eds.), Information Management in Nursing and Health Care, Springhouse Corporation, Pennsylvania, 1996, pp. 74 – 82. [36] C.L. Iacouvou, I. Benbasat, A.S. Dexter, Electronic data interchange and small organizations: adoption and impact of technology, MIS Quarterly 19 (4) (1995) 465 – 485. [37] IEEE Computer Society, IEEE Recommended Practice for Internet Practices-Web Page Engineering-Intranet/Extranet Applications, (1999). [38] H.R. Johnston, S.R. Carrico, Developing capabilities to use information strategically, MIS Quarterly 12 (1) (1988) 36 – 48. [39] D. Jutla, P. Bodorik, J. Dhaliwal, Government support for the e-readiness of small and medium sized enterprises, Proceedings of the 35th Hawaii International Conference on System Sciences, 2002. [40] A. Kambil, A. Kalis, M. Koufaris, H.C. Lucas Jr., Influences on the corporate adoption of Web technology, Communications of the ACM 43 (11) (2000) 264 – 271. [41] H. Kreger, Fulfilling the Web services promise, Communications of the ACM 46 (6) (2003) 29 – 34. [42] A.L. Lederer, A.L. Mendelow, Information resource planning: information systems managers’ difficulty in determining top management’s objectives, MIS Quarterly 13 (3) (1987) 388 – 399. [43] L. Lim, B. Gan, K. Wei, An integrated model on the adoption of internet for commercial purposes, Proceedings of the Thirty-First Hawaii International Conference on System Sciences, 1998, pp. 6 – 9. [44] H. Maruyama, New trends in e-business: from B2B to Web services, New Generation Computing 20 (2002) 125 – 139. [45] H.E. McNay, Enterprise content management: an overview, Proceedings of IEEE International Professional Communication Conference, 2002, pp. 396 – 402. [46] C. Mohan, Dynamic e-business: trends in Web services, in: A. Buchmann, et al., (Eds.), Technologies for E-Services, Third

[47] [48] [49]

[50]

[51]

[52]

[53]

[54]

[55] [56]

[57]

[58]

[59]

[60] [61]

[62]

[63]

[64]

International Workshop, TES 2002, Lecture Notes in Computer Science, vol. 2444, Springer, 2002. M. Naedele, Standards for XML and Web services security, Computer 36 (4) (2003) 96 – 98. R.L. Nolan, Managing the computer resource: a stage hypothesis, Communications of the ACM 16 (7) (1973) 399 – 405. R.L. Nolan, Managing the advanced stages: key research issues, Presentation at the 75th Anniversary Colloquium Series, Division of Research, Harvard Business School, 1983. Oblix, The path to successful Web services: navigating pitfalls, http://itresearch.forbes.com/detail/RES/1087237434_568.html, 2004. G. Pare´, C. Sicotte, Information technology sophistication in health care: an instrument validation study among Canadian hospitals, International Journal of Medical Informatics 63 (2001) 205 – 223. K. Poulter, M. Carter, British American tobacco—in the driver’s seat with Web services-based dashboards, Web Services Journal 4 (1) (2004) 12 – 13. M.S. Raisinghani, Key perspectives on the global e-readiness of websites: a reality check, Journal of Global Information Technology Management 6 (2) (2003) 1 – 5. L. Raymond, G. Pare, Measurement of IT sophistication in small manufacturing businesses, Information Resources Management Journal 5 (2) (1992) 4 – 16. O. Rist, No need to panic yet, InformationWeek 739 (1999 (Jun 21)) 14. M.A. Roth, D.C. Wolfson, J.C. Kleewein, C.J. Nelin, Information integration: a new generation of information technology, IBM Systems Journal 41 (4) (2002) 563 – 577. R. Sabherwal, W.R. King, Decision processes for developing strategic applications of information systems: a contingency approach, Decision Sciences 23 (4) (1992) 917 – 943. R. Sabherwal, P. Kirs, The alignment between organizational critical success factors and information technology capability in academic institutions, Decision Sciences 25 (2) (1994) 301 – 330. D.W. Straub, J.C. Wetherbe, Information technologies for the 1990s: an organizational impact perspective, Communications of the ACM 32 (11) (1989) 1328 – 1399. B.E. Swanson, Information systems innovation among organizations, Management Science 40 (9) (1994) 1069 – 1994. S.L. Talleen, The difference between internet, intranet, and extranet, intranet world, http://www.iorg.com/papers/iw/ 19981019-advisor.html, 1998 (October 19). M. Themistocleous, Z. Irani, A model for adopting enterprise application integration technology, in: D. Bunker, D. Wilson, S. Elliot (Eds.). Proceedings of Working Conference on the Adoption and Diffusion of IT in an Environment of Critical Change (IFIP WG8.6), 61–76 (Sydney, Australia, August 1–3, 2002. A. Tsalgatidou, T. Pilioura, An overview of standards and related technology in Web services, Distributed and Parallel Databases 12 (2002) 135 – 162. J. Tonkel, Flexibility, finances key to Web services adoption, http://www.computerworld.com/developmenttopics/ development/webservices/story/0,10801,95745,00.html, 2004.

A.N.K. Chen et al. / Decision Support Systems 42 (2006) 789–809 [65] USA National Academy of Engineering (Computer Science, Keeping the US Computer Industry Competitive: Systems Integration, National Academy Press, Washington, DC, 1991. [66] J. Ward, P. Griffiths, P. Whitmore, Strategic Planning for Information Systems, Wiley, New York, 1990. [67] L. Wilkes, SOA and Web Services, http://www.looselycoupled. com/opinion/2003/wilkes-soa-infr0415.html, 2003. [68] L.A. Witt, L.A. Burke, Selecting high-performing information technology professionals, Journal of End User Computing 14 (4) (2002) 37 – 50. [69] K. Zhu, K.L. Kraemer, E-commerce metrics for net-enabled organizations: assessing the value of e-commerce to firm performance in the manufacturing sector, Information Systems Research 13 (3) (2002) 275 – 295. [70] G.A. Zolla, Information technology diffusion: a comparative case study of intranet adoption, http://Web.nps.navy.mil/ ~gazolla/files/pp214m.pdf, 1999.

Andrew Chen received his Bachelor of Business Administration from Soochow University at Taiwan, MS in Accountancy from the George Washington University, and PhD in Operations and Information Management from University of Connecticut. His current teaching and research interests include knowledge management, innovation adoption, electronic commerce strategy, database management, and business and Web applications. His research has appeared in Decision Support Systems, European Journal of Operational Research, Journal of Electronic Commerce Research, Journal of Management Information Systems, MIS Quarterly, and international conferences such as AMCIS, DSI, ICIS, and WITS.

809

Sagnika Sen is pursuing her PhD in Information Systems at Arizona State University. She holds an MS in Information Systems and BE in Electrical Engineering. Her research interests include IT service management, service-oriented models of IT, electronic commerce, enterprise systems, and decision support systems. Prior to joining the doctoral program, she worked as an Assistant Systems Engineer in a premier multinational IT services company. Sagnika has presented her work in national conferences like Americas Conference in Information Systems (AMCIS), Hawaii International Conference on Systems Sciences (HICSS), and Workshop in E-business (WEB). Benjamin B. M. Shao received his B.S. in Computer and Information Science and M.S. in Information Management from National Chiao Tung University in Taiwan and his Ph.D. in Management Information Systems from the State University of New York at Buffalo. His research interests include information systems economics, IT security, ecommerce adoption, distributed/parallel processing, and software project management. His research has appeared or is forthcoming in Communications of the ACM, The Computer Journal, Computers & Security, Decision Support Systems, European Journal of Operational Research, IEEE Transactions on Systems, Man, and Cybernetics, Information & Management, Information and Software Technology, Journal of the Association for Information Systems, Journal of Electronic Commerce Research, Journal of the Operational Research Society, and the proceedings of several national and international conferences, among others.